AWOC Winter Weather Track FY06

1. IC6.2 Part 2: Strength and Weaknesses of P-Type

Algorithms
Instructor Notes: Welcome to IC6 lesson 2, Part 2. This lesson is about 30 slides long
and will take approximately 25 minutes to complete.
Student Notes:

2. Outline
Instructor Notes: Lesson 2 of IC 6 is divided into five sections. In Part 1 we discussed
the learning and performance objectives, the utility of p-type algorithms, and then we
focused on the Baldwin technique and the Ramer technique. Now, in Part 2, we will dis-
cuss the Bourgouin Method and Partial Thickness Method.
Student Notes:

3. What about Bourgouin Method?

Instructor Notes: This method was developed by Bourgouin of the Canadian Meteoro-
logical Centre in Dorval, Quebec in the early 1990’s and it has been in use in the Meteo-

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-53

 Warning Decision Training Branch

rological Services of Canada since the mid 1990’s. The method was developed by
looking at a data set of cases in which surface precipitation observations were collocated
with upper air soundings. The upper air soundings provided the vertical temperature
structure needed to establish the criteria utilized in the method. RAOB drift was not
accounted for and values were assumed to be above the sounding site. The precipitation
had to be reported within one hour of the time that the sounding was taken and at the
same location for it to be included in the data set utilized to tune the method. This criteria
and the fact that the technique was tuned over only two winter seasons (1989-90 and
1990-91) resulted in a small training data set. For example, it was initially tuned to 54
cases of freezing rain versus ice pellets, 119 cases of rain versus snow, and 3-5 cases of
ice pellets versus rain. The number of cases with respect to ice pellets versus rain is
uncertain in that there is no information on the exact state of the hydrometeor aloft, (i.e.,
as it enters a surface based warm layer) so the number of cases is estimated.
Student Notes:

4. Bourgouin Method – the Basics

Instructor Notes: The method is built on the premise that the temperature variation of a
falling hydrometeor and its resulting phase changes are predominately driven by the
temperature of the environment through which the hydrometeor falls. Clearly, a hydrome-
teor falling through a layer with temperatures above freezing temperatures will become
liquid if it resides in this layer for a sufficient time. Given this premise, the method
assumes that two parameters drive p-type, the mean temperature of a layer and the res-
idence time of the hydrometeor in this layer. The mean temperature of a given layer can
be readily computed for both observed and forecast soundings. The same cannot be
said of the residence time of the hydrometeor in a layer as the residence time is depen-
dant on the depth of the layer, the terminal velocity of the hydrometeors, and the mean
vertical motion. Therefore, the method assumes a constant vertical motion and terminal
velocity. The assumption of a constant terminal velocity could be considered a weakness
as different crystal structures have different falls speeds. However, it is not clear how sig-
nificant this variation could be or how one could account for it operationally.

6-54 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

Student Notes:

5. Bourgouin Method – the Basics

Instructor Notes: Given these concepts it is clear that one can combine the mean tem-
perature of a layer and the height of a layer by multiplying them together which will give
you a value proportional to the area covered by the layer in question. Essentially, the
algorithm determines if there is sufficient energy to melt or freeze hydrometeors as they
fall. A positive area is defined as the area between the 0 degree C isotherm and the envi-
ronmental temperature in the above freezing layer. Conversely, a negative area is
defined as the area between the 0 degree C isotherm and the environmental tempera-
ture in the below freezing layer. Positive and negative areas in a given sounding are
computed, bounded by the 0 degree C isotherm and the environmental temperature
curve. The precipitation type (snow, freezing rain, ice pellets, or rain) is determined by
the magnitude of the positive and negative areas. In cases where two precipitation types
are determined to be of equal likelihood, one or the other is chosen randomly.
Student Notes:

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-55

 Warning Decision Training Branch

6. Bourgouin Method – Rain vs. Snow

Instructor Notes: The most obvious forecast is when ice crystals are present and the
entire profile is sub-freezing then you will have snow or the entire profile is above freez-
ing in which case you will obviously have rain. Obviously, rain versus snow consider-
ations get much more complex than this. When you have a surface based warm layer (or
positive area) with no warm air aloft (in other words all negative area aloft) the possibili-
ties are confined to rain, snow, or a mix of the two. To differentiate between profiles of
this type the Bourgouin method interrogates the positive area at the surface.
Student Notes:

7. Bourgouin Method – Rain vs. Snow

Instructor Notes: In cases such as the one we just described, the table on the right is
utilized to differentiate between rain and snow. If the surface based positive area is less
than roughly 5.5 J/kg it will predict snow. If the surface based positive area is greater
than a little over 13 J/kg it is determined to be rain. Between these two values the
method assumes a rain/snow mix. Here is an example of how the snow versus rain table
looks when displayed in the Bufkit software. Note they only considered cases with small
surface based positive areas (less than 20 J/kg) as surface warm layers stronger than
this were associated with surface temperatures well above freezing and were clearly
strong enough to completely melt any frozen hydrometeors. Also, it is important to note
that the Bourgouin method assumes ice crystals are entering the surface layer. It does
not perform a check for the initial state of the hydrometeor.

6-56 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

Student Notes:

8. Assumption of Ice Crystals?

Instructor Notes: The Bourgouin Method does not check the negative area aloft for sat-
uration to ensure that the moisture is deep enough and temperature cold enough to sup-
port heterogeneous nucleation of ice crystals. This is the primary weakness of the
method and one that you should keep in mind when you look at output from the Bour-
gouin method. Note in the example shown that the saturated layer only extends to a tem-
perature of -4 or -5 degrees C and that the profile is very dry at temperatures of -10
degrees C and below. Clearly this is a case where it would be unlikely that we would
have ice crystals present. However, the output from the Bourgouin algorithm indicates
snow. Utilizing the top-down methodology you will quickly catch this, but if you were to
look at the algorithm output in a vacuum you may get a forecast of snow in what turns out
to be a freezing drizzle event. BUFKIT has an added check to account for this and will
display a message (as seen on this slide) alerting you to the likelihood that heteroge-
neous nucleation may not occur in the given profile.
Student Notes:

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-57

 Warning Decision Training Branch

9. More on Bourgouin Method

Instructor Notes: The database utilized to tune the Bourgouin method did not contain,
or only contained a very small number of, cases of mixed snow and freezing rain or
mixed snow and ice pellets, so these scenarios are not addressed by the method. Based
on the soundings studied during development of the algorithm it was found that it only
took a positive area of 2 J/kg to induce the transition from frozen to liquid precipitation. If
the positive area aloft (if any) is less than 2 J/kg then it assumes that the ice crystals are
not melted and it will look at the low levels to test the rain versus snow criteria. If there is
a positive area of 2 J/kg or greater aloft then the algorithm assumes complete melting of
the ice crystals and then it will look at the low level to determine if the resultant precipita-
tion type should freezing rain, ice pellets, or rain.
Student Notes:

10. Bourgouin Method – ZR vs. IP

Instructor Notes: This example demonstrates how the Bourgouin method would con-
sider a freezing rain or ice pellet case. It was found in the research that the most promis-
ing predictors of ice pellet versus freezing rain cases were the positive and negative
areas in the sounding (that is a positive area above a surface based negative area). The
positive area aloft would be enough to melt any frozen hydrometeors, in this example. If
the hydrometeor completely melts and then falls into a sub-freezing surface layer then
the question becomes is there enough negative area to completely re-freeze the hydrom-
eteors as ice pellets or will it reach the surface as freezing rain.

6-58 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

Student Notes:

11. Bourgouin Method – ZR vs. IP

Instructor Notes: The Bourgouin study actually found a pretty clear delineation
between freezing rain and ice pellets in scenarios such as this based on the strength of
the positive area aloft and the negative area near the surface. They found that even a
small positive area aloft (say, greater than 2 J/kg) will lead to freezing rain so long as the
surface based negative area is relatively small (say, less than 50 J/kg). For cases with a
weak positive area (left hand side of the image) it was found that a negative area greater
than around 100 J/kg would lead to ice pellets while cases with stronger positive areas
aloft would require a stronger negative area to completely refreeze the hydrometeors to
ice pellets. The study noted that with a small positive area (2 J/kg or less), it is possible to
get some snow mixed in with the freezing rain or ice pellets. However, they did not
attempt to discriminate whether or not freezing rain or ice pellets would be mixed with
snow. Obviously, if you have a reliable sounding near these thresholds you will likely
want to forecast a mix of precipitation types. The resultant diagram shown on this image
is utilized in the Bourgouin method to differentiate between ice pellets and freezing rain
when you have a positive area aloft and a negative area at the surface.
Student Notes:

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-59

 Warning Decision Training Branch

12. Using the Bourgouin Method

Instructor Notes: The area (or Bourgouin) method follows a theoretical hydrometeor as
it falls through different temperature layers. In the sounding you will note that there are
two positive areas (one at the surface and one aloft) with a negative area in between the
two. To determine the p-type in a scenario such as this the method will attempt to mimic
the behavior of the hydrometeor as it descends. Let’s look at how the Bourgouin tech-
nique would handle a scenario such as this. The method assumes that the hydrometeors
start off as ice crystals, so the first question is whether the positive area aloft is sufficient
to melt the ice crystals. As the crystal enters the positive area aloft the method would
assume complete melting as this positive area is greater the 2 J/kg. So, as it exits the
positive area aloft the method would treat hydrometeors as being in liquid form. It would
then check to determine if the negative area is strong enough to completely re-freeze the
hydrometeors by looking the strength of the positive area aloft and the negative area
beneath it. If the negative area is less than 75-100 J/kg it will treat the hydrometeors as
liquid as they exit the bottom of the negative area. Then as the hydrometeors enter the
surface based positive layer, in this case, the method will determine the resultant precip-
itation type as rain.
Student Notes:

13. Using Bourgouin Method - BUFKIT

Instructor Notes: Keep in mind that the appearance of the Bourgouin charts will
change depending on the vertical temperature profile. Depending on the vertical distribu-
tion of the positive and negative areas different charts are utilized to assess the resultant
p-type. In this first example, we have two positive areas adjacent to an elevated negative
area. The Bourgouin method treats this as a sleet versus rain decision and utilizes the
chart shown on the left. In this next example, we have a positive area aloft and a nega-
tive area at the surface. This is treated as a freezing rain versus sleet scenario. Finally, in
this last example we have a negative area aloft and a positive area at the surface. This is
a rain versus snow decision and the Bourgouin method utilizes the chart displayed on the
left. Keep in mind that as you step through time the vertical profile will change. As the

6-60 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

vertical profile changes the distribution of the positive and negative areas will change as
well which may result in different Bourgouin charts being utilized for different time periods
in the forecast time frame of concern. The best place to view the output from the Bour-
gouin technique is in the BUFKIT software. In BUFKIT the Bourgouin output can be dis-
played for any model sounding. As you step through the hourly or three hourly
soundings, with the Bourgouin technique toggled on, the appropriate chart will display
based on the distribution of the positive and negative areas in the profile.
Student Notes:

14. Bourgouin Technique: Strengths and Weaknesses

Instructor Notes: The Bourgouin technique has several strengths. First, it was devel-
oped based on observed soundings and the precipitation type associated with those
soundings. It can also be applied to any region or model output. The technique also has
a high POD for freezing rain cases. There are also several weaknesses to the method.
Foremost of these is the assumption that ice crystals are present. The assumption that
heterogeneous nucleation has occurred can easily lead to precipitation type forecast fail-
ures. Another weakness is that it does not account for dry layers. This has the same
effect as what was discussed with the Ramer technique in that hydrometeors are not
allowed to sublime. This is compounded by the methodology’s use of the temperature as
opposed to the wet-bulb temperature. By utilizing the temperature and ignoring dry lay-
ers the technique does not appropriately consider sublimation or evaporation and the
associated wet-bulb cooling as hydrometeors fall through an unsaturated layer. Finally,
the technique assumes a constant terminal velocity of hydrometeors. Hydrometeors
have different terminal velocities based on their size and shape and this impacts the res-
idence time of a hydrometeor in a given layer.

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-61

 Warning Decision Training Branch

Student Notes:

15. Quiz Break 1

Instructor Notes: Time for a short Quiz Break. Assume that you have precipitation
associated with this sounding. Which of the Bourgouin technique’s weaknesses would
come into play with this profile? When you are ready to see the answer click ‘Next’ to
advance to the next slide.
Student Notes:

16. Quiz Break 1

Instructor Notes: The correct answer is C. Given that the profile is saturated to only -5
degrees C the Bourgouin technique’s assumption that ice crystals are present would
come into play in this example. The Bourgouin would assume that ice crystals are
present and since there is no positive area in the sounding it would forecast a precipita-
tion type of snow. Since the sounding is only saturated to around -5 degrees C it is
unlikely that ice crystals are present in this cloud layer and the most likely precipitation
type would actually be freezing rain.

6-62 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

Student Notes:

17. What About the Partial Thickness Technique?

Instructor Notes: The partial thickness method was developed prior to widespread
access to gridded data and BUFR soundings. Essentially, the method assesses layer
thickness values to interrogate warm and cold wedges aloft and at the surface. While the
technique is not as detailed as more rigorous sounding analysis methods such as the
top-down methodology it can be useful in plan view mode as it may assist the forecaster
in the identification of areas that require more thorough inspection. This methodology is
also useful in later time periods when you are looking to get handle on the potential
range in the precipitation type and not as confident in the details of the vertical profile.
Student Notes:

18. Partial Thickness Technique – the Basics

Instructor Notes: Forecasters have been using thickness to forecast precipitation type
for many years. However, recent advances in computing power have allowed forecasters
and researchers to develop and apply some new approaches to this technique. Thick-
ness techniques are based on the fact that the thickness of a layer, defined by 2 constant
pressure levels, is proportional to the temperature of the layer, with lower thicknesses

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-63

 Warning Decision Training Branch

corresponding to lower temperatures. Previous research has shown that reliable precipi-
tation type thresholds for thicknesses can be determined for a given station. For exam-
ple, it was determined many years ago that snow was favored over rain at many
locations when the 1000-500 mb thickness is below 540 dm. Similar thresholds have
been determined at many locations for smaller layers, such as 1000 to 850 mb and the
850 to 700 mb. The term “Partial” in Partial Thickness refers to the fact that these layers
comprise only part of the traditional 1000 to 500 mb layer. Modern computing power now
allows forecasters to easily compute observed and model forecast thicknesses for a vari-
ety of layers, and allows forecasters to combine and display combinations of various
thicknesses in a variety of ways in order to visualize the horizontal and vertical tempera-
ture profile in an effort to forecast precipitation type.
Student Notes:

19. Partial Thickness Thresholds

Instructor Notes: The thickness thresholds indicated by the nomograms shown on this
slide were developed for Greensboro, North Carolina. Research in the Carolinas indicate
that these thresholds work well at locations with elevations ranging from sea-level to
around 1500 feet above sea-level. Note that the forecast technique suggested by these
nomograms involves combining low-level thickness with mid-level thickness to derive
precipitation type. For example, a small 1000-850 mb thickness in combination with a
large 850-700 mb thickness (the lower-right portion of the nomogram) implies cold air at
low levels, with warmer air aloft, and the likelihood of freezing precipitation.

6-64 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

Student Notes:

20. Visualization of the Location and Orientation of P-

type Boundaries
Instructor Notes: One advantage of utilizing the partial thickness method in the fore-
cast process is the increased ability to visualize the location and movement of precipita-
tion type boundaries across wide areas in plan view. In the example shown on this slide,
the plot of partial thickness values would help the forecaster visualize the likely location
of precipitation type boundaries in a complex situation. The forecaster would then be
able to further refine his or her forecast by performing a high resolution sounding analy-
sis at a few key locations utilizing a more rigorous methodology such as the top-down
method. In addition, viewing this type of display in a time-lapse mode would also allow
forecasters to visualize the evolution of the atmospheric thermal environment; for exam-
ple, the advance or retreat of cold air at low or midlevels.
Student Notes:

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-65

 Warning Decision Training Branch

21. Visualization of Atmospheric Processes

Instructor Notes: Another application of partial thickness is shown on this slide. View-
ing the time-evolution of partial thickness values at a particular site can give the fore-
caster insight into what kind of atmospheric processes are at work. For example, in the
graph at the upper left, low-level temperature advection, in the absence of mid-level tem-
perature advection, is indicated when the low-level thickness is changing, but the mid-
level thickness remains constant. The graph at the upper right shows the opposite condi-
tion; mid-level temperature advections are indicated with an absence of low-level tem-
perature advection, when the mid-level thickness values change while the low-level
thickness values remain constant. The 2 graphs at the bottom of this slide show the
response of the thickness values in sample situations for various combinations of lower
and mid-level temperature advection.
Student Notes:

22. Partial Thickness Technique: Strengths and

Weaknesses
Instructor Notes: In summary, the use of partial thickness to forecast precipitation type
has both strengths and weaknesses. The first strength is that partial thickness values are
derived from fields, specifically height fields, that are well-forecast by the models. In con-
trast, sounding interrogation techniques sometimes rely on accurate model forecasts of
shallow warm layers, and subtle boundary layer temperature structures, that may not be
as well-forecast. Another strength is that partial thicknesses can be displayed in ways
that help the forecaster visualize the location and movement of precipitation type bound-
aries, and to visualize the atmospheric processes that are affecting the vertical profile.
The primary weakness of the partial thickness technique is that it does not take advan-
tage of the full vertical resolution of the model. For example, the NAM model currently
has a vertical resolution of 60 layers, with the majority of those layers located below 500
mb. Partial thickness techniques only involve the diagnosis of a very limited number of
those layers (typically 2), and may sometimes miss important details in the forecast verti-

6-66 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

cal structure of the atmosphere, especially in situations where very shallow layers of
warm or cold air are present. For example, rain can sometimes occur at stations near
large, unfrozen bodies of water when partial thickness values are below typical rain/snow
thresholds, if a shallow, mild layer of maritime air is present near the surface. Another
weakness of partial thickness is that threshold values can vary by station, depending on
the geographic characteristics of the station. For example, locations at higher elevations
tend to have higher rain/snow thresholds of partial thickness. Forecasters should also
keep in mind that microphysical processes are not accounted for in any partial thickness
technique. For example, the partial thickness technique may erroneously predict snow in
situations where the entire sounding is below freezing, yet the moist layer does not
extend into a layer where the temperature is below -10 degrees C and therefore the pro-
file is not likely to support heterogeneous nucleation. Because of these weaknesses, it is
not recommended that partial thickness forecasting techniques be used alone to forecast
precipitation type. If you choose to utilize the partial thickness technique it may be best to
utilize partial thicknesses for plan view visualization, and visualization of atmospheric
processes, and then utilize more through sounding interrogation to account for shallow
warm and cold layers, as well as for microphysical considerations.
Student Notes:

23. Quiz Break #2

Instructor Notes: Time for a short quiz break. Which of these selections is one of the
primary weaknesses of the partial thickness method? Click ‘next’ to advance to the next
slide.

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-67

 Warning Decision Training Branch

Student Notes:

24. Quiz Break #2

Instructor Notes: The correct answer is A. One of the weaknesses of the partial thick-
ness technique is that it does not take advantage of the full vertical resolution of current
numerical models.
Student Notes:

25. Summary: Strengths and Weaknesses of

Algorithms
Instructor Notes: One key aspect to take away from this presentation is that all p-type
algorithms have individual strengths and weaknesses. The best way to determine precip-
itation type is to utilize your own knowledge of the environment and potential model
errors while applying the top-down methodology. However, p-type algorithms can assist
you in the process of making the determination of p-type, particularly when you are look-
ing at many points and time periods. The Baldwin method is widely available and is easy
to understand and visualize why it is producing a given precipitation type. Another
strength is it's use of the wet-bulb temperature to determine precipitation type. The pri-

6-68 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

mary weakness of the Baldwin method is with how it handles deep isothermal layers
near the surface with temperatures between 0 degrees C and -4 degrees C. If these lay-
ers are sufficiently deep the Baldwin method will produce a forecast of rain or freezing
rain it what is ultimately a snow sounding. The Ramer method may be the strongest algo-
rithm statistically speaking. Another strength is its use of the wet-bulb temperature in a
top-down process to determine precipitation type. It may be a more robust algorithm than
the others, but it is correspondingly more difficult to understand how it arrived at a given
precipitation type. While it is not as visually appealing as the others the primary weak-
ness of the method is it's inability to assess dry layers. The Bourgouin method again is
easy to understand and has very useful charts to help visualize why it is producing a
given precipitation type. The method is based on precipitation types associated with
actual observed soundings which is a strength. However, the Bourgouin method does
not check the initial state of the hydrometeor and, as a result, will assume ice crystals are
present in some situations where heterogeneous nucleation has not been achieved.
Another significant weakness is the use of T instead of Tw. Partial thickness concepts
are particularly useful in plan views and for visualization of the potential distribution of
precipitation. The methodology itself may not be as detailed as the others, but it readily
lends itself assessing a larger area without requiring us as forecasters to look at a multi-
tude of point soundings. Significant weaknesses of partial thickness techniques are that
they do not consider microphysical concepts and do not take advantage of the details
available in high resolution model data.
Student Notes:

26. Availability of Different Techniques

Instructor Notes: Here is a brief review on the availability of the different algorithms.
The Baldwin technique is available for the NAM BUFR data in BUFKIT and the NAM pre-
cipitation type selections in the AWIPS browser. It is also available on the EMC web site
which shows an ensemble of precipitation type algorithm output. The Ramer technique
is available for the GFS BUFR data in BUFKIT and is also available with the LAPS pre-
cipitation type display in AWIPS. The Ramer output is also displayed on the EMC web
site alongside the Baldwin technique output. The Bourgouin method output can be dis-
played for any BUFR data output in BUFKIT. However, it is not available in AWIPS at this

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-69

 Warning Decision Training Branch

time. Output from the partial thickness method can be displayed for any BUFR output in
BUFKIT. It is also available in AWIPS as a local add-on to the Volume Browser selec-
tions.
Student Notes:

27. Interactive Quiz

Instructor Notes: Take a few moments to complete this interactive quiz.
Student Notes:

28. Questions???
Instructor Notes: After going through this lesson if you have any questions, first ask
your SOO, or local AWOC winter track facilitator. Your SOO or local facilitator should be
able to help answer many questions. If you need additional info from what your SOO pro-
vided, send an E-mail to the address on the slide. This address sends the message to all
the instructors involved with this IC. Our answer will be CC’d to your SOO so that they
can answer any similar questions that come up in the future. We may also consider the
question and answer for our FAQ page. Thanks for your time and good luck on the exam!

6-70 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms

 AWOC Winter Weather Track FY06

Student Notes:

29. References
Instructor Notes: References
Student Notes:

30. References
Instructor Notes: Additional references

IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms 6-71

 Warning Decision Training Branch

Student Notes:

6-72 IC6.2 Part 2: Strength and Weaknesses of P-Type Algorithms